PARTIAL OXIDATION OF FOSSIL AND RENEWABLE FUELS preprint

17th Int. Symp. on Plasma Chemistry, August 7-12, 2005, Toronto, Canada

Partial oxidation of fossil and renewable fuels into the Synthesis Gas A. Czernichowski, M. Czernichowski, P. Czernichowski1 ECP – GlidArc Technologies, La Ferté St Aubin, France

Abstract The mixture of CO and H2 is produced in our GlidArc (a High-Voltage discharge) assisted reformers via partial oxidation of various fuels with atmospheric air. Such process accepts almost any feed at up to 4% of Sulphur content. The feed conversion is total. No soot, coke or tars are produced. The assisting electric power presents less than 1% with respect to the power of the processed fuel. Keywords: Synthesis Gas, Syngas, GlidArc, Reforming, Partial Oxidation 1. Introduction The Synthesis Gas (SynGas, a mixture of mostly CO and H2 ) is widely used as source of Hydrogen (via a simple extraction or when CO is water-shifted to CO2 and H2 ), as feedstock for chemical syntheses or as gaseous fuel for power generation. The last application can be performed, for example, in the Solid Oxide Fuel Cells (SOFC) that are capable of operating directly with the SynGas containing light Hydrocarbons and even some H2 S. The SOFC operating at high temperatures allow also a heat recycling for the primary fuel processing or other uses. Conversion of liquid fuels into gaseous reformate for SOFC feeding is however difficult. High Sulphur and aromatic content of certain fuels deactivates classical catalytic processors and/or creates a significant soot appearance. Instead of removing thousands of Sulphur-organic molecules from fuels we propose a very simple plasma-assisted conversion of almost any fuels into the SynGas containing all initial Sulphur converted into the Hydrogen Sulphide. Such SynGas can be then very easily cleaned from H2 S, if necessary. Our High-Voltage discharge called GlidArc [1] assists first the cold-start of such processors when the full combustion of fuel is performed at high turbulence conditions. Once the reformer is sufficiently hot the discharge starts to assist only partial oxidation of the fuels with a reduced amount of oxidant. Such assistance allows the feed pre-reforming in the plasma zone so that the process achievement in the post-plasma zone containing an activated solid material may occur. Our electrically assisted process of partial oxidation accepts almost any feed at up to 4% of Sulphur content. The feed conversion is total. No water (or steam) needs to be added so that the process is simple. Our tests with natural gas, cyclohexane, heptane, toluene, gasoline and Diesel oil (DO) were first published in [2]. Then we have presented a study on propane (LPG) [3] and natural gas [4]. The Rapeseed oil (Canola) conversion into SynGas was then tested [5] as example of various renewable fuels processing. 2. GlidArc-assisted reformers Some experiments are now presented with a military aviation fuel and Rapeseed oil. We return also to our previous studies on the Diesel oil reforming and show our recent results. Here-presented tests are performed in small or medium-size reactors. A 0.6-L (internal volume) reformer is shown on the Fig. 1. The plasma zone contains two flat electrodes that delimit a space filled with the gliding discharges. A feed + preheated air are mixed, blown into that space and flows along the diverging electrodes. The discharges ionize the air + partially vaporised fuel. Given the moderate temperature of the electrodes (not cooled) and a very short contact time of the discharge roots with the electrodes, we do not observe any deterioration (even at a high Sulphur presence) that may prevent the gliding of these current-limited discharges. Fig. 1. Schematic of a 0.6-L GlidArc-assisted reformer. 1

Injector of Fuel + Air

Electrodes GlidArc

Post -plasma zone

SynGas

Thermocouples

presently with Ceramatec Inc.

1


The plasma zone communicates with a post-plasma zone filled with activated packing (noble, rare or exotic elements are not used for activation). The flow of partially converted reactants containing long-living active radicals enters the zone where the conversion is completed by deactivation of all excited species. A 5 to 10kV power supply provides both ionisation of the air/fuel mix and then a transfer of the electric energy into the plasma. The time-averaged electric power is measured at the mains; it is less than 0.1kW for small reformers or up to 0.4kW for large ones. The reformer is thermally insulated to keep it as hot as possible. Its total inside volume is 0.6L (1.8 to 6.6L for 20 or 80kW processor, the output power being accounted as the Lower Heating Value LHV of exiting reformate gas). No part of reactors is cooled in a forced manner. Some thermocouples measure the postplasma zone temperatures. The output gas sample crosses white wool for soot presence checking. Other sample is analysed using a two-channel µ-GC dedicated to H2 , N2 , O2 , CH4 , and CO for one channel, and CO2 , C2 H4 , C2 H6 , C2 H2 , C3 H8 +C3 H6 , and residual moisture for the second one. Figure 2 presents a picture of 6.6-L reformer that we are currently testing for up-to 1kg/h Hydrogen (+ roughly 14kg CO) production from the Soybean biodiesel or fossil Diesel oil. Such SynGas (25m3 (n) per hour equivalent to 80kW LHV) is devoted to clean up a NOx-polluted exhaust from a 2MW Diesel engine. The GlidArc-assistance power is only 0.4kW.

Fig. 2. 6-L GlidArc-assisted reformer of various fuels.

3. Reforming 3.1.Aviation fuel The JP-8 military aviation fuel can be considered as a logistic support for SOFC-based auxiliary power supply. The fuel has the following characteristics: relative density 0.800, average formula CH1.94, aromatics 15.3 vol.%, total Sulphur 433ppm wt., initial BP 150°C, final BP 252°C, net heat of combustion (LHV) 43.2MJ/kg. The explored ranges of inputs were: JP-8 @ H2 10–16 4.4–8.3g/min and air @ 23–41L(n)/min. The CO 15–20 post-plasma zone of the processor contained H2 +CO 6–35 0.34L of activated refractory granules. The N2 +Ar 5–68 temperature of exiting SynGas (Reformate) CO2 2.2–4.8 reaches 725°C. As the result of various runs CH4 0.7–2.8 we have always observed the absence of soot, C2 H4 0.01–2.8 tar or non-reacted feed in the Reformate gas. C2 H6 0.00–0.19 The dry gaseous output (in vol.%) was: C2 H2 , O2 , C3+, others absent From the mass and enthalpy balances we deduce the output Reformate gas power (based on the LHV of all combustible components acceptable for SOFC): it can reach 4.4kW. Knowing the net heat of combustion (LHV) of the JP-8 we calculate the thermal efficiency of the JP-8 conversion into Reformate gas in our processor. This efficiency (defined as the ratio of the LHV of H2 +CO+light hydrocarbons at 25°C to the

2


20

CO H2 CO2 CH4 C2H4

10

%

vol.% (dry)

15

80

8

60

6

40

4

20

kW

LHV of the entering JP-8 fuel) is up to 78% for explored range of input parameters. Remaining energy content (chemical enthalpy) of the initial fuel is converted into the heat – but it is a compromise that we propose to accept for such a simple GlidArc-assisted soot-free and Sulphur-resistant technology. The total GlidArc assistance power varied from 0.13kW at the cold-start of the process to some 0.05kW at the stabilized reforming conditions. The process power consumption related to the output Reformate gas power presents therefore only 1%. Figs. 3 and 4 show a typical composition of the Reformate gas (dry) and give an idea on the reformer output dynamics and its thermal efficiency.

2

5

thermal efficiency output kW LHV 0

0 4

0 4

5

6 7 JP-8 input g/min

8

9

Fig. 3. Composition of the Reformate gas as a function of JP-8 fuel input rate.

6

8

JP-8 input g/min

Fig. 4. Reformate output power and the thermal efficiency of the reforming as a function of JP-8 fuel input rate.

A 12-hours run at the constant air flow rate of 29L(n)/min and the constant fuel flow rate of 6g/min of JP-8 in 0.6-L processor at 3kW (LHV) output Reformate gas power and 60W GlidArc assistance was successfully preformed. No structural changes of the processor and its post-plasma zone after the run were found. Neither soot nor tars were produced. We installed therefore this reformer for online feeding of a SOFC stack in a laboratory. Four series of tests were performed. No problems appeared after long runs during which several litres of this high Sulphur fuel crossed the reformer. Our 79 gas analyses showed a stable Reformate gas composition and the mass balance closure was typically 1%.The SOFC stacks were composed of a single 10-cell and then a dual 11-cell system. The stacks worked perfectly. Their performance were comparable to the Hydrogen baseline and the electric power output difference correlated well with higher steam content in our reformate gas. No soot was found in manifolds or electrodes after the stacks dismantling. 3.2. Diesel oil Already in 2001-2 our feasibility tests with various road Diesel oils were successful [2] and further tests with a highly Sulphur-polluted heavy oil showed us that practically no harm is observed up to 4% Sulphur content. We performed then more detailed studies and long tests expanding the output Reformate gas power to >20-kW scale. Our 1.8-L reformer is very similar to the previously described one. The plasma compartment (0.6L) contains two electrodes powered through one 10-kV power supply. The same activated material is used in the postplasma zone of 1.2L volume. French road DO (from Total) is processed. Its average Carbon number is 15.6 (ranging from C8 to C29 ); an averaged formula can be written as CH1.83, and the molecular mass is in the range of 210–220. This fuel has a quite high relative density of 0.826 and a high Sulphur content of 310ppm by weight. The DO (dosed by a metering pump) and compressed air (controlled by a mass flow meter) are simply mixed in a "T" connector and preheated up to 140-200°C before their injection to the GlidArc zone by an 8-mm (inner diameter) tube centred on the electrode axis. Through a porthole one can see the plasma discharge in the stream of the air + DO droplets. The droplets do not however affect our electric discharge. Explored inputs are: Compressed air 48–146L(n)/min Preheat 140–200°C Diesel Oil 11–30g/min GlidArc power 0.3–0.4 kW As result we observe the following outputs at no sooting conditions and at total fuel conversion: Bottom reformer temperature: ≤870°C

3


25

kW otput

20 15 10 5 0 10

20

30

40

g/min DO input

Fig. 5. LHV power output of 1.8-L processor as a function of the input flow rate of Diesel oil.

50 45 40 35 30 25 20 15 (dry) vol.% 10 5 0 0.45

0 1 2 3 4 5 6 7 8 9 10 kWh per kg DO

Reformate gas content (dry basis, vol.%): H2 16–20 CO 19–22 H2 +CO 38–41 CO2 2.4–4.8 CH4 0.8–3.3 C2 H4 0.0–2.1 N2 +Ar 52–58 C2 H2 , O2 absent Reformate gas LHV power: 7-22kW. Fig. 5 and 6 show the LHV power output of all combustibles (H2 , CO, CH4 , and C2 H4 ) as a function of the DO input flow rate. They also illustrate the specific LHV energy output (in kWh per kg of DO) and the dry Reformate gas composition as a function of O2 /C molar ratio at the input (reflecting the air/fuel ratio).

CO2 C2H4 CH4 H2+CO kWh/kg

0.5

0.55

0.6

0.65

O2/C molar ratio

Fig. 6. Specific energy (LHV) output and the Reformate gas composition as a function of O2 /C at the input (right).

Our tests are successful: such quite highly Sulphur polluted Diesel oil is totally reformed at no water or steam added and at no soot or tars production (when checking the gas output with a white wool). To confirm that absence we dismantled several times our processor after its cooling and did not find any soot deposits inside. Our uninterrupted 24-hours test showed a good stability of the system. Other long-term tests are under way… 3.3. Rapeseed oil (Canola) We use other 1-L reactor, similar to that shown on the Fig. 1. This processor contains three electrodes connected to 3-phase power supply 100 mA current to each electrode. No preheat is applied to any stream. Several runs were preformed using an edible oil having the density of 913kg/m3 . At the accumulated runtime of 6 hours we never changed any part of the reformer. The process was always very stable. When soot was starting to appear at an insufficient air/oil ratio we just added more air (or reduced the oil flow) to establish the non-sooting reforming conditions. During these runs the air flow rate was comprised between 35 and 102 L(n)/min for the oil input flow rate between 11 and 30mL/min. Fig. 7 shows temperatures T1 (upstream) and T2 (downstream, equal to exiting gas product temperature) such as observed when the thermal equilibrium of the reactor is achieved. Various tests are put together on the same graph showing therefore the explored range of the parameters. Fig. 8 presents the composition of the output gas (all successful tests are put into the same graph). In fact, the SynGas is always diluted by Nitrogen as we use air for reforming. The N2 concentration at the output is typically in the 55–60 vol.% range (dry basis) depending on the air/oil ratio applied for the given test. As results of the mass balances we calculate the output flow rate of 100% SynGas (H2 +CO only) for all runs as shown on the Fig. 9. The H2 +CO flow rate in the output can also be presented as the potential thermal output power. Such a power is therefore assimilated to the LHV. Our runs and reactor are not yet optimized but they indicate that we already obtain (at atmospheric pressure) as much as 11kW or 3.5m3 (n)/h of pure SynGas output. Higher outputs are expected for elevated pressures and/or preheat of the air and oil streams.

4


3000

1200

2000

70 60 CO2

50 vol.%

1500

L(n)/L

4000

900

H2

40

N2

30

CO

20

1000

600

T1 T2

air/oil ratio L(n)/L

0 10

20

10 0 3000

300 30

3500

4000

air/oil L(n)/L

rapeseed oil input mL/min

Fig. 8. Concentration of the main components (dry basis) in the output gas as a function of the air/oil ratio for all successful runs. Other gases are at minor concentrations: CH4 0.5–1.0, C2 H4 0.1–0.5, C2 H6 0.01– 0.03, and C2 H2 0.001–0.005 vol.%.

12

72

kW L(n)/min

10

60

8

48

6

36

4

24

2

12

0

L(n)/min of SynGas

output LHV power (kW) of SynGas

Fig. 7. Temperatures T1 (upstream) and T2 (downstream) inside the post-plasma zone as a function of the input flow rate of rapeseed oil. The left axis presents the air/oil ratio required for completely nonsooting reforming of the oil.

0 10

15

20 oil mL/min

25

30

Fig. 9. Output flow rate of 100% SynGas (H2 +CO only) in L(n)/min and in kW for all runs at close-to the thermal conditions of the reformer.

The thermal efficiency of the reforming process can be estimated from these results. If one takes the LHV of the feed as 35MJ per kg then for our run at 30.4mL/min = 0.507mL/s = 0.463g/s the thermal input power corresponds to 16kW. For 11.4kW thermal power (LHV) output that we obtain as SynGas one finds therefore the thermal efficiency of the process at no sooting conditions equal to 70%. This quite high efficiency is based on only standard enthalpy heats (at 25°C) of the output H2 +CO combustion to CO2 and steam (LHV) with respect to the standard heat of combustion (LHV) of entering oil. Such efficiency does not take into account any residual light hydrocarbon gas useful for SOFC or the sensitive heat content in the gas leaving the reformer. A part of that heat can certainly be reused to preheat entering feed and air so the thermal efficiency could be increased. No such heat exchange was used in this study. 3.4. Other feeds We have already performed positive tests of reforming using 90% Ethanol, Glycerol, saturated Sugar/water solution, Soybean oil, Soybean diesel, and heavy bio-oil from the flash pyrolysis of wood. All that opens opportunities to upgrade some farm-issued products or waste biooils from various activities. ECP is presently fabricating a 30-L processor of about 0.5 MW output LHV power. Some last-minute results will be presented during the Symposium. 4. Conclusions • Working hours or days we never changed any part of the reformer and/or its post-plasma zone loading. The technology is robust.

5


•

Reformate gas appears after ~15min from a cold start or after ~2min when the reformer is kept hot. At such quick starts we do not obtain yet optimal performances – but the Reformate gas flow may be sufficient to start operate a SOFC (that also asks for a quite long heating and equilibration period). • Thermal efficiency of the process is around 75%. An overall efficiency of our fuel processor + SOFC system can be increased thanks to some synergetic effects. • Some methane and ethylene are present in the Reformate gas. They can be kept at quite low concentration, if necessary, by applying a higher O2 /C ratio at the input - but it would slightly lower the thermal efficiency of the reforming. These gases are however considered as good fuel for SOFC. • Reformer works also when Sulphur is present in the feed. We proved it when reforming the JP-8 and Diesel oil at respectively 433ppm and 310ppm (weight) content. Our previous experiments show that even 4% of Sulphur in a heavy liquid carbonaceous feed (end distillation point of 600°C) is not harmful. In fact, even 100% H2 S can be processed using our GlidArc reactors for Hydrogen recovery from such waste gas [6]. • Conversion of the fuels is total, as we do not find any residual fuel or Oxygen at the exit. We do not produce soot, coke or tar and that is at no water or steam addition. • Noble, rare or exotic elements are not used for activation of the solid mater present in the post-plasma zone of the reformer. • Process is stable. The Reformate gas output flow and composition can be kept constant or match a required level by regulating the fuel and air flow rates. Drastic changes of the Reformate gas output flow and/or composition can be done in a fraction of minute. • Electric assistance is low, around or less than 1% relative to the LHV of produced Reformate gas flow. • According to our tests we can produce a flow of Reformate gas that is equivalent to 12kW LHV power per one Litre of inside reformer volume. Here presented results show a simple way for a Reformate gas production from fossil or renewable matter. Our past tests show that pure Oxygen or O2 -enriched air can be used instead of atmospheric air. It opens some opportunities for more efficient GlidArc-assisted reformers and processes if such extra Oxygen is applied to reduce the Nitrogen content in the Reformate gas… One can also reduce the reformer volume or increase its output when working at higher pressures; we have some results at up to 6 bars. References [1] A. Czernichowski, M. Czernichowski, Further development of plasma sources: The GlidArc-III, this Symposium. [2] A. Czernichowski, M. Czernichowski, P. Czernichowski, Non-catalytical reforming of various fuels into syngas, France-Deutschland Fuel Cell Conf. on "Materials, Engineering, Systems, Applications", Forbach, France, 2002, p. 322-8. [3] A. Czernichowski, M. Czernichowski, P. Czernichowski, GlidArc-assisted reforming of various carbonaceous feedstocks into synthesis gas. Detailed study of propane reforming, 14-th Annual U.S. Hydrogen Meeting, 2003, Washington, DC, CD-proceeding, 8 pp. [4] A. Czernichowski, K. Wesolowska, GlidArc-assisted production of synthesis gas through partial oxidation of natural gas, First International Conference on Fuel Cell Science, Engineering and Technology, Rochester, NY, April 21-23, 2003, p. 181-5. [5] A. Czernichowski, M. Czernichowski, K. Wesolowska, GlidArc-assisted production of synthesis gas from Rapeseed oil, Hydrogen and Fuel Cells Conf. and Trade Show, Vancouver, Canada, 2003, postconference proceedings (CD), 6 pp. [6] A. Czernichowski, P. Czernichowski, M. Czernichowski, GlidArc-assisted removal and/or upgrading of Hydrogen Sulfide or Methyl-Mercaptan, 16th Int. Symp. on Plasma Chemistry, Taormina, Italy, June 22-27, 2003, poster, CD proceedings, 7 pp.

6